How Much Should You Raise? Fund the Proof, Not the Fantasy

Quibi raised $1.75 billion before it had proved that people would pay for short-form mobile video. Nubank raised $2 million to prove a much narrower idea: an app-managed credit card in Brazil. Their contrasting stories reveal the real rule for sizing a startup round. The target should cover the work needed to reach a milestone that the next capital provider can verify, plus the time and cost of raising again. It should also fit the kind of capital available once that proof exists.

Quibi had money before it had an audience.

The US streaming company raised $1.75 billion before launch, committed about $1 billion to first-year content and mounted a reported $470 million marketing campaign. It launched on 6 April 2020. Six months later, it announced it was closing.

COVID-19 damaged Quibi’s commute-focused idea. But the more revealing problem came earlier. The company had already built a blockbuster cost structure before it had shown that people would pay for its short-form mobile video, share it or make it part of their lives.

That raises an uncomfortable question for every founder: can too much money arrive too early?

The number that looks reassuring

Founders are often handed a shortcut. Carta’s Q2 2026 median US seed round was $4.5 million. It is a useful reference point. It is not a funding plan.

For an AI-enabled software team, $4.5 million may fund years of unnecessary hiring and expansion. For a regulated hardware company, it may not even get the product through testing and manufacturing. One benchmark can be excessive and inadequate at the same time. Finance has a gift for making this sound less obvious than it is.

The useful question is not, “What do seed-stage companies raise?” It is: “What must we prove before the next credible source of capital will take us seriously?”

Then comes the part founders regularly undercount: you need cash not merely to reach that proof point, but to verify it, prepare the next raise and close it.

Carta reported that the median company raising a Series A in Q4 2024 had waited 774 days, or 2.1 years, from its prior round. It now advises founders to plan for 24 to 30 months rather than 12 to 18. The milestone may happen on schedule. The money after it may not.

Nubank chose the smaller fight

In 2013, David Vélez raised a $2 million seed round from Sequoia and Kaszek for Nubank, the Brazilian financial technology company that would eventually challenge the country’s banks.

He did not set out to fund a new bank immediately. Nubank focused on an app-managed credit card, one of the few financial products it could initially offer without becoming a regulated bank. Its first card transaction took place on 1 April 2014. When the company launched publicly that September, it raised approximately $14.3 million to $15 million.

Nubank used a narrowly defined card product to create trust and underwriting evidence before pursuing broader ambitions. Photo: Nubank / Wikimedia Commons, Public domain.

At first glance, this sounds like a story about caution. It was really a story about sequencing. The card gave Nubank something more valuable than a grand ambition: customer trust and underwriting data. It created evidence for the next decision.

That is the central insight. A well-sized round buys a fact that the next funder can check.

For venture investors, that fact may be a working product, customer adoption, recurring revenue, regulatory clearance or sound unit economics, meaning whether each sale makes or loses money. Y Combinator’s practitioner rule is to finance the next “fundable” milestone while generally avoiding more than 25% seed dilution, the share of the company handed to new investors.

GI Network’s view: The best round size is not the number that makes a pitch look ambitious. It is the number whose final pound, dollar or euro produces evidence that changes the next capital decision.

The cost of getting to “yes”

The operating calculation is simple, even if the spreadsheet rarely is:

Raise = cumulative net cash burn through the milestone and next close + one-off milestone costs + contingency − cash available − committed non-dilutive inflows.

Net cash burn means the cash leaving the company each month after cash received. Non-dilutive inflows are funds, such as grants, that do not require founders to give up ownership.

Imagine a software company with $50,000 in monthly net burn for 24 months, $100,000 in one-off costs, a 15% contingency and $200,000 already in the bank. The illustrative calculation produces a $1.295 million requirement. At an $8 million pre-money valuation, meaning the agreed value before the new money enters, that means roughly 14% dilution.

Raise Carta’s $4.5 million median instead and the dilution in this example becomes 36%.

The interesting part is not that 36% is always wrong. It is that money changes behaviour. A bigger bank balance can bring senior hires, new markets and fixed commitments that make the next milestone more expensive to achieve. Research by Alex Murray and Greg Fisher, published in *Organization Science* in 2022, found that broad, unbounded venture claims attracted more resources but were associated with lower long-term viability through technological complexity and uncontrolled scaling. It is association, not proof that large rounds cause failure. Still, Quibi makes the mechanism painfully easy to picture.

Read how much company you should give up and who gets paid first before treating dilution as a harmless headline percentage.

What investors are really buying

A founder sees a round as runway. An investor sees a claim on an uncertain future.

Ownership matters because the investor needs enough of the eventual upside for a successful outcome to compensate for the many companies that do not succeed. That is why a fund may care about how much of the company it owns after the seed round, not just the size of its cheque.

Follow-on reserves matter too. These are funds an investor holds back to invest again in companies that prove themselves. A seed investor is often not deciding only whether to support today’s company. It is deciding whether the company can reach a later round with a story strong enough to justify putting in more money.

And then there is downside protection. 54gene, the Nigerian genomics company, accepted a down round reportedly carrying a three-to-four-times liquidation preference after its finances deteriorated. A liquidation preference determines who gets paid first if the company is sold or wound down. It can protect new investors, but it can sharply change what remains for founders and earlier shareholders.

This is where most people stop looking. A bigger round can reduce immediate fundraising pressure while increasing ownership loss today and the complexity of future financing tomorrow.

The capital source changes when the proof changes

Freshworks, the Indian software company then known as Freshdesk, did not begin with a fashionable institutional round. It signed its first customer in June 2011 and reached 100 customers in 103 days. It won $40,000 from Microsoft’s BizSpark competition before Accel invested $1 million in December 2011. A $5 million round followed in April 2012.

Founder Girish Mathrubootham later advised founders to begin pitching after building something meaningful and generating user traction. Freshworks did not need to prove every part of its future. It needed to prove that customers would pay.

M-KOPA, the Kenya-based provider of financed products and services, shows the next stage of the same logic. It had 300 customers in 2011. By April 2023, it had served more than three million customers and delivered $1 billion in cumulative products and services. Its May 2023 financing exceeded $250 million, including more than $200 million of sustainability-linked debt led by Standard Bank and $36.5 million in equity from Sumitomo.

M-KOPA shifted towards debt once customer repayments had created a proven asset base lenders could assess. Photo: Nicholas Githiri / Pexels, Pexels licence (free commercial use).

The company’s customer repayments had become an asset base lenders could assess. That changed the appropriate capital source. Debt, borrowed money that must be repaid, could finance much of the expansion without selling more ownership.

SVB’s venture-debt guidance illustrates why this is not emergency money. It generally looks for at least $4 million raised in one equity round, reputable venture backing, demonstrable growth, more than 12 months of organic runway and a clear milestone plan. Typical venture debt is 20% to 40% of the last equity round and is intended to add roughly six months, not revive a company with an almost-empty account.

For more on that distinction, see why bankable ARR is a different test.

When more money early is the sensible answer

The Quibi lesson has limits.

A larger early round can be justified when the next fundable proof cannot be reached in cheap stages. Consider the brief’s illustrative regulated-hardware company: it burns $200,000 a month for 30 months, needs $2 million for testing and manufacturing, carries a 20% contingency and has $500,000 in cash. It requires $9.1 million. A $4.5 million benchmark seed would leave a $4.6 million completion gap.

Here, underfunding can be more reckless than raising big. Testing, equipment and production work may be irreversible technical or regulatory costs. Stopping halfway does not create a smaller, elegant version of success. It creates an unfinished programme.

The European Innovation Council recognises that reality. Its fund supports companies at technology-readiness levels 5 to 9, uses due diligence and releases investment through milestone-based tranches, with follow-on decisions as progress is demonstrated. The point is not to deny the company enough capital. It is to release capital against work that can be checked.

That is a better answer than pretending every deep-technology business should behave like a software startup. It may also point to grants, strategic capital or staged financing. Read why early infrastructure should be funded through evidence first if physical capacity is part of the next milestone.

Demand can make the problem worse

Zano, the nano-drone project from UK company Torquing Group, sought £125,000 through crowdfunding. It raised about $3.4 million from roughly 15,000 orders.

That looked like validation. It was not manufacturing readiness.

Kickstarter’s commissioned investigation concluded that the campaign presentation was misleading about technical readiness and that the team lacked the experience and resources to manufacture the promised product. More orders meant more tooling, components, fulfilment and support obligations before production had been validated.

Quibi bought scale before proving demand. Zano received demand before proving scale. Different routes. Same trap.

54gene reveals a third version. After COVID-testing success, the Nigerian genomics company expanded into diagnostics through Seven River Labs, hired more than 100 people and acquired expensive equipment. The company had raised $45 million across three rounds since its 2019 founding. Reported expected returns did not materialise. By mid-2022 it had run short of cash, cut salaries and staff, and began winding down in July 2023.

54gene’s expansion illustrates why exceptional revenue should not automatically fund permanent cost structures. Photo: Artem Podrez / Pexels, Pexels licence (free commercial use).

Put the three cases side by side and a pattern appears that none of the reports states outright: capital is dangerous when it turns an unverified signal into a permanent obligation. Quibi had attention and content. Zano had orders. 54gene had temporary revenue. None, on its own, justified the cost structure that followed.

What founders should do before naming a number

Start by writing one sentence: “When we achieve this result, this type of funder can verify it and finance our next step.”

Then cost every month until that proof is achieved and the next round is closed. Separate one-off testing, regulatory, equipment and manufacturing costs from ordinary operating spend. Add contingency where uncertainty is real. Subtract only cash already available and funding already committed.

Next, identify the irreversible commitments. A permanent hire, laboratory equipment, a lease or a production contract should each answer one question: does this directly produce the next fundable proof point?

Finally, match the request to the capital source. UK Start Up Loans require a business plan, a 12-month cash-flow forecast, a personal budget and three months of bank statements. They offer £500 to £25,000 at 7.5% fixed interest, repayable over one to five years. That is useful for the right business. It is not venture capital in miniature.

GI Network would test the funding plan before outreach: identify the precise milestone, challenge the burn and completion assumptions, assess whether future cash flows can support debt, map equity, grant and strategic-capital options, and rehearse the ownership and downside questions an investment committee will ask.

The Proof, Pause, Pay framework

Before setting your target, use three questions.

Proof: What single result must the company produce next, and who can independently verify it?

Pause: Which costs become difficult to reverse before that result is proved, and can they be delayed or broken into stages?

Pay: Who should fund the next stage, what do they need to see, and what are you giving them in return through ownership, repayment obligations or downside protection?

If the answers fit together, you have a financing plan. If the number came first, you probably have a fundraising aspiration.